Beyond Starry Night: Shortcut-Aware Control-State Planning for Artist-Grounded Text to Image Generation

2026-08-06

AI Summary

Current text-to-image models often rely on superficial shortcuts—such as fixed color palettes or recurring motifs—when generating images in a specified artist’s style, failing to faithfully capture both the intended scene and authentic artistic characteristics. To address this, this work proposes Atelier, a novel framework that introduces, for the first time, a shortcut-aware control state planning mechanism. Atelier explicitly models scene anchors, style hypotheses, preserve-or-transform decisions, artist-specific evidence, and anti-shortcut constraints, integrating artist-knowledge-guided state construction, local patch referencing, and iterative global-local feedback optimization. The authors also introduce ArtIntentBench, a new benchmark for systematically evaluating style fidelity and shortcut avoidance. Experiments demonstrate that Atelier significantly enhances style faithfulness across both open- and closed-source models, better preserves original structural content, and effectively suppresses stylistic shortcuts, outperforming baselines based on prompt engineering, retrieval augmentation, and general-purpose agents.

Abstract

Artist-grounded image generation requires more than appending an artist name to a prompt. Image models often respond to artist names through canonical shortcuts, such as recurring motifs, generic palettes, or overrepresented period signatures, rather than preserving the user's intended scene. We introduce Atelier, a shortcut-aware control-state planning framework for artist-grounded image generation. Atelier translates underspecified artistic intent into an explicit control state that separates scene anchors, preserve/transform decisions, style-regime hypotheses, role-bound artist evidence, and shortcut-avoidance constraints. It grounds this state using artist-level knowledge and local patch references, compiles backend-aware generation plans, and iteratively refines candidates through global and local authenticity feedback. We further introduce ArtIntentBench, a benchmark covering Van Gogh and Qi Baishi across artwork re-rendering, period/style-controlled generation, historically unseen subjects, shortcut auditing, and human preference evaluation. Across open-weight and closed-source generators, Atelier improves artist-level style fidelity, preserves source structure more faithfully, and substantially reduces shortcut substitution compared with prompt-engineered, retrieval-augmented, and general-purpose agent baselines. These results suggest that artist-grounded generation is bottlenecked not only by image synthesis, but by the upstream inference of explicit, evidence-grounded artistic controls.

Problem

• artist-grounded image generation
• shortcut learning
• style fidelity
• text-to-image generation
• artistic intent

Innovation

• shortcut-aware planning
• artist-grounded generation
• control-state representation
• style fidelity
• evidence-grounded control